Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?

The DSE site says the stick is USB2.0.
My USB external HDD can read @ ~28Mbyte/s thru the same USB ports.

The file copy starts off with a ~4min estimate, and then blows out to
50min, oscillating about.

Performance monitor showed a steady baseline write speed of
200KBytes/s, peaking to 800KBytes/s sometimes and 6MBytes/s rarely.
A disk test utility showed it can be read from @ 12MBytes/s.

Windows 2000 Server SP4, patched.

Oh well...

Mark

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

"Mark C" <mark.c@somewhere.invalid> wrote in message
news:47e0db8b$0$28141$c3e8da3@news.astraweb.com...
> Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
>
> The DSE site says the stick is USB2.0.
> My USB external HDD can read @ ~28Mbyte/s thru the same USB ports.
>
> The file copy starts off with a ~4min estimate, and then blows out to
> 50min, oscillating about.
>
> Performance monitor showed a steady baseline write speed of
> 200KBytes/s, peaking to 800KBytes/s sometimes and 6MBytes/s rarely.
> A disk test utility showed it can be read from @ 12MBytes/s.
>
> Windows 2000 Server SP4, patched.
>
> Oh well...
>


Are you transferring many, many files? And is your USB drive formatted Fat16
by any chance? If so, that could be the source of your trouble.

http://blogs.technet.com/markrussino...1/2087460.aspx

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

On 19 Mar 2008 09:23:23 GMT, Mark C <mark.c@somewhere.invalid> wrote:

>Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
>
>The DSE site says the stick is USB2.0.
>My USB external HDD can read @ ~28Mbyte/s thru the same USB ports.
>
>The file copy starts off with a ~4min estimate, and then blows out to
>50min, oscillating about.
>
>Performance monitor showed a steady baseline write speed of
>200KBytes/s, peaking to 800KBytes/s sometimes and 6MBytes/s rarely.
>A disk test utility showed it can be read from @ 12MBytes/s.
>
>Windows 2000 Server SP4, patched.

Try a different USB port on the machine.

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

impossible wrote:
> Are you transferring many, many files? And is your USB drive formatted Fat16
> by any chance? If so, that could be the source of your trouble.
>
> http://blogs.technet.com/markrussino...1/2087460.aspx

Large numbers of files alone will also dramatically slow file copying to
flash memory, although it may depend on how you have it configured. I
have mine configured for performance over reliability.

Flash memory is inefficient at wriging, so having to constantly update
file tables with large numbers of entries will be much slower than
copying a single file. If I need to pur a large number of files on flash
memory, I will usually zip them up first, and it is much faster that way.

RL

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

In article <frufks$jvb$1@lust.ihug.co.nz>, RL did write:

> Large numbers of files alone will also dramatically slow file copying to
> flash memory ...

I don't see why. I've had directories with hundreds of thousands of files in
them, which could be copied around just fine between rotating media, across
the LAN etc. Why should flash be any worse at dealing with this?

> ... although it may depend on how you have it configured.

How do you do that, then?

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

Lawrence D'Oliveiro wrote:
> In article <frufks$jvb$1@lust.ihug.co.nz>, RL did write:
>
>> Large numbers of files alone will also dramatically slow file copying to
>> flash memory ...
>
> I don't see why. I've had directories with hundreds of thousands of files in
> them, which could be copied around just fine between rotating media, across
> the LAN etc. Why should flash be any worse at dealing with this?

This is because of the way flash memory works. Memory is arranged in
blocks, of which wach bit has a natural value of 1. The state of the bit
can be individually changed from 1 to 0, but not 0 to 1. Therefore, to
write a small block of data as is required for a file table entry, the
entire block must be erased and re-written.

>> ... although it may depend on how you have it configured.
>
> How do you do that, then?

On Windows, this is set in the device properties. In Vista, this is
accessible via Explorer, right click on a drive and select Properties.
Select the Hardware tab, then the flash device from the "All disk
drives" list. Select 'Properties', then the 'Policies' tab. You get two
options "Optimize for quick removal" and "Optimize for performance".

RL

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

In article <frunas$ofg$1@lust.ihug.co.nz>, RL did write:

> Lawrence D'Oliveiro wrote:
>
>> In article <frufks$jvb$1@lust.ihug.co.nz>, RL did write:
>>
>>> Large numbers of files alone will also dramatically slow file copying to
>>> flash memory ...
>>
>> I don't see why. I've had directories with hundreds of thousands of files
>> in them, which could be copied around just fine between rotating media,
>> across the LAN etc. Why should flash be any worse at dealing with this?
>
> This is because of the way flash memory works. Memory is arranged in
> blocks, of which wach bit has a natural value of 1. The state of the bit
> can be individually changed from 1 to 0, but not 0 to 1. Therefore, to
> write a small block of data as is required for a file table entry, the
> entire block must be erased and re-written.

What do you mean by "file table entry"? Do you mean "inode"? And what
difference does it make how big or small they are, when you can just write
a bunch of them at once?

>>> ... although it may depend on how you have it configured.
>>
>> How do you do that, then?
>
> On Windows, this is set in the device properties. In Vista, this is
> accessible via Explorer, right click on a drive and select Properties.
> Select the Hardware tab, then the flash device from the "All disk
> drives" list. Select 'Properties', then the 'Policies' tab. You get two
> options "Optimize for quick removal" and "Optimize for performance".

Sounds like a long-winded way of setting the "cache size".

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

"RL" <rl@null.void.test> wrote in message
news:frunas$ofg$1@lust.ihug.co.nz...
> Lawrence D'Oliveiro wrote:
>> In article <frufks$jvb$1@lust.ihug.co.nz>, RL did write:
>>
>>> Large numbers of files alone will also dramatically slow file copying to
>>> flash memory ...
>>
>> I don't see why. I've had directories with hundreds of thousands of files
>> in
>> them, which could be copied around just fine between rotating media,
>> across
>> the LAN etc. Why should flash be any worse at dealing with this?
>
> This is because of the way flash memory works. Memory is arranged in
> blocks, of which wach bit has a natural value of 1. The state of the bit
> can be individually changed from 1 to 0, but not 0 to 1. Therefore, to
> write a small block of data as is required for a file table entry, the
> entire block must be erased and re-written.
>
>>> ... although it may depend on how you have it configured.
>>
>> How do you do that, then?
>
> On Windows, this is set in the device properties. In Vista, this is
> accessible via Explorer, right click on a drive and select Properties.
> Select the Hardware tab, then the flash device from the "All disk drives"
> list. Select 'Properties', then the 'Policies' tab. You get two options
> "Optimize for quick removal" and "Optimize for performance".
>
> RL
This is just setting write behind caching. Turned on you get windows back
quicker and writing continues in the background. This may lead you to
believe save is complete and pull out stick. Disabled, means saving data
ties up the app until finished. Doesn't stop you pulling it out early but at
least you have a indication that data is still saving.

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

On Fri, 21 Mar 2008 11:09:46 +1300, RL <rl@null.void.test> wrote:

>Lawrence D'Oliveiro wrote:
>> In article <frufks$jvb$1@lust.ihug.co.nz>, RL did write:
>>
>>> Large numbers of files alone will also dramatically slow file copying to
>>> flash memory ...
>>
>> I don't see why. I've had directories with hundreds of thousands of files in
>> them, which could be copied around just fine between rotating media, across
>> the LAN etc. Why should flash be any worse at dealing with this?
>
>This is because of the way flash memory works. Memory is arranged in
>blocks, of which wach bit has a natural value of 1. The state of the bit
>can be individually changed from 1 to 0, but not 0 to 1. Therefore, to
>write a small block of data as is required for a file table entry, the
>entire block must be erased and re-written.
>
>>> ... although it may depend on how you have it configured.
>>
>> How do you do that, then?
>
>On Windows, this is set in the device properties. In Vista, this is
>accessible via Explorer, right click on a drive and select Properties.
>Select the Hardware tab, then the flash device from the "All disk
>drives" list. Select 'Properties', then the 'Policies' tab. You get two
>options "Optimize for quick removal" and "Optimize for performance".

So if you've got it optimised for performance then you'd have to make
sure you always stop the device so it's safe to remove?

Re: Should it take 50 minutes to copy 1.5GB to a USB2 memory stick?
 

In article <1t88u3p1kd2lm8e80gm55fl3ue54d7rioj@4ax.com>, Craig Shore did
write:

> So if you've got it optimised for performance then you'd have to make
> sure you always stop the device so it's safe to remove?

That would be exactly right. "Performance" means "caching", and "caching"
means "risk of data loss if caches are not properly flushed".